AI processes reduce radiation dose during medical treatments
Medical engineer Prof. Christoph Hoeschen, who holds the Chair of "Medical Technology Systems" at the Faculty of Electrical Engineering and Information Technology, was presented with an award from the German government by Parliamentary State Secretary at the Federal Ministry for the Environment, Dr. Bettina Hoffmann, for the success of a lighthouse project he led in research into AI in medical imaging.
In the joint research project KI-INSPIRE, OVGU, the German Cancer Research Center (DKFZ) in Heidelberg and the University of Lübeck developed procedures that could reduce the radiation dose in interventional medical imaging by two-thirds using AI and safely test these procedures. "Radiation-based procedures such as computed tomography are important in diagnostics and minimally invasive therapy in order to obtain a picture of medical conditions and treat patients appropriately," emphasizes Prof. Hoeschen. "However, this also involves a high level of radiation exposure, both for patients and for medical staff."
Hoeschen and his team, together with the partners in their research project, have used AI to improve image quality while using less raw data. "During the intervention, the same areas are repeatedly recorded in the image and a needle or catheter is inserted at the same time. This means that there is a lot of preliminary information. And I can have this information integrated into the subsequent images by an AI. Then I don't need live images and therefore have a lower radiation dose. The AI algorithms also filter out interference when processing the raw data. This results in images with a higher image sharpness, less noise and fewer artifacts. This can be done live, the data is virtually provided online and the doctor could see this directly on the screen in the future."
This requires various further steps: Firstly, consideration must be given to the potential implementation of the procedures in interventional imaging systems that generate computed tomographic images, as the algorithms can only be used sensibly if they are used directly in clinical systems. On the other hand, further research is needed to really prove that the procedures provide at least as good information for the treatment of patients as previous procedures, although the exposure to ionizing radiation could be so much lower. The quality assurance of the procedures and their interaction with the hardware has also not yet been solved. To this end, the 3D-printed phantoms developed in the project, which contain structures similar to human tissue, are to be further developed.
The federal government is currently considering further funding measures to further promote the implementation of the results of the lighthouse projects.
The KI-INSPIRE project was a three-year so-called AI lighthouse project of the Federal Government, which was carried out jointly by the Chair of Medical Technology Systems at OVGU, the CT working group at the DKFZ led by Prof. Dr. Marc Kachelrieß and the Chair of Instrumentation of Medical Imaging at the University of Lübeck led by Prof. Dr. Magdalena Rafecas. The OVGU working group coordinated the project.
The AI lighthouse projects are advertised by the Federal Government by topic by various ministries in order to promote AI-based processes and make Germany more competitive in this area.